Difference between revisions of "Audio distribution"
|Line 270:||Line 270:|
Revision as of 03:02, 11 November 2008
Sometimes its desired to distribute an audio feed around the house. This makes it possible to listen to the radio or a recording wherever you go, without carrying equipment around. It also makes it easy to listen to and record from the stereo when that and the computer are in different locations, without the inconvenience and downsides of digital distribution. Other uses are also discussed.
At first sight it might appear that the installation of home computer networking would make an audio feed redundant, but this isn't really so. While computer networking can be used to distribute audio, there are significant issues with this approach:
- The resulting audio is not always properly synchronised between computers
- Sound is only available where computers are located, making coverage patchy
- Computer reception & playing isn't automatic, making such a system inconvenient
- Often some of the networked computers are switched off, giving worse coverage & more inconvenience
- wireless data links give poorer synchronisation and sometimes breakup
- Adding another sound receiver requires a whole computer system
The good news is now that people are installing wired computer networks, audio distribution can be installed at the same time with almost no extra work or cost, making physically distributing audio pretty simple. This enables this or various other applications to be used later if desired. See Low Voltage Wiring for more info.
The hum issue
The main potential issue with analogue audio distribution is hum. There are various ways to avoid hum infecting the audio signal, and 2 simple options are presented here.
The simplest option is to use screened wire for the feed. The cable's outer conductor, the screen, connects to the chassis of the source stereo. This screen shields the inner conductor from hum pickup. Distributing stereo requires 2x screened wires, which is available as a single figure of 8 shaped twin screened cable if preferred. Twinax could also be used, but is less convenient to terminate and costs more.
However this approach means buying a reel of screened cable, when one could instead use cat5e, which you already have when networking, is much cheaper, and which can also carry other signals for other jobs.
The other possible issue with screened is that if a long length of wire is fed from old audio equipment with high line output impedance, high frequency loss is likely, and hum posible.
When using this approach, the cable needs to be fully screened. There are screened cables on the market that are very light on screening copper, resulting in a partially screened cable. These are not suitable.
Low impedance feed
This option works with any type of wire, and cat5e is generally the most convenient. No screening is used, instead the signal is driven from a low impedance source. This is most easily achieved by feeding the audio output to a small audio amplifier (eg 1 watt), and connecting the output of that to the cable. Such amplifiers have output impedances of under an ohm*, and this low impedance eliminates hum.
* note an amplifier's real output impedance is not the same as its intended load impedance. (And its not unusual to see one described as the other.)
Headphones can be run directly from a low impedance feed. Simply fit a headhone socket, and plug in. Headphones use TRS sockets in 2 sizes, either 3.5mm or 1/4".
In principle there's a limit to how many headphones can be run at once, and this limit depends on the characteristics of both headphones and feed amplifier. However in practice it would be unusual to encounter a domestic setup where the feed amp's abilities were exceeded.
Where its wanted to use small compact active speakers, such as computer speakers, distributing the power required to run them down the same cat5e can in principle avoid the need for a number of local wallwarts, or the need to run wires from the speakers to mains sockets.
However there are significant problems amd gotchas with this approach. If you don't understand them or aren't willing to spend a little time debugging then this topology is better avoided.
Note this option is only for low impedance audio feeds, its not suitable for high impedance screened feeds. The increase of Power over Ethernet (PoE) for Voice over Internet Protocol (VoIP) telephones may make streaming audio appliances powered from the Ethernet line viable in future.
A fully reliable bug free system can be realised if the 2 points below are followed.
First, the audio ground wire must not be used to distribute the power ground, 2 separate conductors are required for the power distribution.
Secondly the active speakers should have a differential input to ensure good behaviour. This is more likely to be found on speakers intended to run from a USB connection; speakers that come with their own wallwart normally don't have a differential input.
There is no sure way to tell differential input amps from ones that aren't, unless you have electronics skills and can examine the circuitry. But a quick test can tell in some cases. Checking the resistance from input ground to power ground (on the amp or active speaker):
- if the 2 are connected, you have an amp without diff input
- if they aren't connected (near zero resistance), the test is inconclusive
Speakers without differential input
Most computer speakers don't have differential input. These will work fine when each is run off its own local wallwart. Running such speakers from a single central supply is a recipe for trouble, though it can be done.
Speakers without differential input on distributed power
Its possible to use speakers without differential input and feed them all from a central power source, but with this combination problems are likely, debugging time is necessary, and even after solving them the setup is not ideal.
If its not known whether speakers have a differential input, they can be tried. If they don't have it, a problem may occur, which shows up as unwanted noises on the sound, typically plopping, farting, growling, motorboating, or squealing noises. This problem is most likely to happen at maximum volume, so turn them up to check.
There are various possible solutions to parasitic noises described below, but the use of non-differential input speakers with power distribution always runs the risk of affecting sound quality at low frequencies, and while this often won't be noticed with computer speakers, it may be noticeable when listening through a hifi system. So if you want to ensure best sound quality, avoid any hassle and you use non-differential input speakers, then each such speaker should be fed from its own wallwart rather than using distributed power.
To obtain optimum performance for recording on this system, switch all the fed speakers off during recording time.
If using speakers without differential input and feeding them all from a central power source, there are several methods to reduce power/audio feed interaction and the risk of noises it causes. Whether such a topology causes any low frequency distortion depends on how thoroughly the problem is tackled, and the problem is not always 100% solvable. However its quite practical to try to run a system this way, knowing that if issues should persist one can simply revert to using a local wallwart for each speaker to solve it.
Problem noises can often be eliminated by connecting a large capacitor across the speaker's power connections, 2200uF or larger. If some active speakers on the system exhibit a problem and some don't, its best to connect a capacitor to every speaker. This minimises the chance of one speaker affecting another. Capacitors should be used with all systems using this topology. If 2200uF doesn't fix it, try a 4700uF cap with a 0.1uF capacitor (preferably ceramic) in parallel with it. (Note such capacitors are polarised, and need to be connected + to the positive wire, - to the negative.) The 3rd option, a bit more effective again, is to use the following CRC filter (the diode behaves as a non-linear resistor):
+ o----+-----|>|---+-----+----o + to speaker
| | | C1 D C2 C3 | | |
- o----+-----------+-----+----o - to speaker
C1 = 2200uF C2 = 2200uF C3 = 0.1uF D = 1A diode, eg 1n4001
If capacitors don't eliminate it the problem, another option is to parallel conductors for the negative power feed wire. The downside is this then uses 3 or 4 wires instead of 2. Another possibility is to fit bell wire or speaker wire when installing network wiring, and use this thicker bell wire to distribute the speaker power. This will much reduce power/signal interaction. (Bell wire has a small 3 dimensional stripe on one side used to tell + from -, you need to look closely to get the polarity right at each termination.) Another option is to fit an additional single wire during installation, and use this for the power feed ground wire.
Positioning the power supply used for the feed in the geographic centre of the network minimises the length (and hence resistance) of cable from power supply to speakers, and also reduces the effect of one speaker on another.
Using an unearthed power supply for the feed reduces interaction further. Generally plastic cased supplies are unearthed, metal cased are earthed.
Where one speaker is on a long run compared to the others, it will be much more prone to problems than the others, plus each active speaker contributes to some extent to the risk of trouble with the others. One option is to either use a local wallwart for that one or use a passive speaker. A passive speaker may give sufficient volume, but sometimes it may not, as the audio power is limited by the power output of the feed amplifier, and further restricted by the resistance of the cable run.
If in spite of these measures, parasitic noise or distortion at low frequencies occurs, then lose the power feed and use local wallwarts on each speaker, and it should all behave perfectly.
In practice there are areas covered by sound from more than one speaker or pair of speakers. If both are not in phase with each other, partial loss of bass occurs, and less even sound distribution. It isn't necessary to phase adjacent sets of speakers, but doing so does improve the evenness and quality of the sound.
When speakers are out of phase, the problem is solved by either reversing the phase of one set or by pointing them in the opposite direction.
Phase reveral is easy with passive speakers (just swap the wires over) but not so easy with active speakers.
Speakers with differential inputs normally work fine if the audio input wires are swapped over.
Non differential input speakers can have the audio input wires swapped if powered by local wallwarts, and usually work ok.
Never swap input wires over with:
- non differential input speakers if powered from a central power feed
- any earthed appliance, such as a computer, tv, stereo system, etc
If one of these is done, the speaker may not work, and other problems can sometimes occur.
For non differential input speakers powered from a central power feed there is only one way to reverse phase: open the speakers up and swap the 2 wires going to the speaker driver. Pointing the speaker in the opposite direction is often easier.
- Active speaker
- a speaker with built in ampifier. These require a source of power. Nearly all modern computer speakers are active.
- Differential input
- an amplifier (which may be built into an active speaker) that responds to the difference between the 2 audio wire inputs. This has a few advantages.
- Distribution line
- aka distribution cable, feed line etc. The cable that runs round the house to distribute the signal.
- Passive speaker
- Speaker with no built in amplifier. These don't require any power source, just the audio signal itself. Small computer speakers from the win3.1 era were mostly passive. Hi-fi speakers are normally passive.
- S/N ratio
- The ratio of wanted audio signal to unwanted noise. The noise usually consists of hiss, hum & distortion products, hopefully at very low levels.
No audio system is perfect, but for recording, a few tips can help this analogue approach get as close to perfect as equipment will allow.
Run the feed amplifier at around half maximum volume. This optimises S/N ratio while maintaining good headroom. Max volume is just under the point at which the amp begins to distort, and is not determined by volume knob position.
Its best not to route your network wiring with mains wiring.
When using non differential input speakers on central power distribution, its best to switch the central power feed off when recording, as if left on the speakers can affect sound quality in many cases.
When using multiple feed amplifiers driving one distribution line, switching off any feed amps not being used maximises S/N ratio.
External wire runs
Exterior wiring is vulnerable to physical damage, and should be routed where this isn't likely. Wiring at or below ground level benefits from physical protection, and the simplest option for underground signal wires is to thead it through garden hose.
Wiring mounted above ground is vulnerable to UV damage, and for longest possible life it should either be
- shaded from the sun
- or be painted with oil based gloss paint (topcoat only)
To avoid exterior wiring faults affecting the system or frying the feed amp, the exterior cable is best fed via a pair of resistors (eg 15 ohm). These go in the L+ and R+ lines.
Very long wire runs
For very long wire runs, eg covering several houses or very large buildings, either of the following works:
1. Use co-ax wire for distribution, driving it with a feed amplifier.
2. Use a bridged output feed amplifier, which gives a low impedance balanced output, and receive the signals with an amplifier with differential inputs. Note that this requires 4 feed conductors, the two -s must not be connected, and non differential input equipment and 3 terminal connectors must not be connected to the feed line. Consequently it may be convenient to arrange the system with a balanced main feed driving several smaller local unbalanced feeds to cover the various service areas.
What plugs in where
The 'record output' (may be labelled 'tape out') from the source (usually a stereo amplifier) is connected to the input of the feed amp using a standard screened lead.
The feed amp's speaker output terminals connect to the distribution line.
- the L wire in the line connects to the L speaker + terminal
- the R wire in the line connects to the R speaker + terminal
- the ground wire in the line connects to - terminal of either speaker connection on the feed amp.
- Also connecting a 2nd ground wire to the other speaker - terminal makes connecting plugs & sockets to the distribution wiring easier, and reduces problems when using central power feed, but its not essential.
Distribution sockets have 3 or 4 terminals. A resistor goes between the feed line's L+ and the socket's L+, and another resistor goes between the feed line's R+ and the socket's R+. Distribution line's ground wire(s) connect to the socket's ground connection(s).
When more than one feed amp is used, the feed line's L+ is connected to the feed amp L+ through a resistor, and the feed line's R+ is connected to the feed amp R+ through a resistor. Feed line ground goes to one of the speaker - terminals on the feed amp.
Can't I just connect to the stereo's speaker output?
If you do it will work, but all the adjustments the amplifier makes will then affect the sound feed. These include bass & treble controls, volume and balance, and any other sound controls fitted. This is fine for casual listening, but for recording audio on the computer its a real curse. Its also not ideal in that speaker resonances can affect the signal quality.
If you want you can run 2 feeds in the one cat5 cable:
- from the main amp output, which feeds standalone speakers, controlling volume etc house-wide from the source stereo system.
- from the small feed amplifier - this feed is used for recording.
This would use a minimum of 5 of the 8 wires in a cat5 cable.
3.5mm TRS jacks allow computers, headphones or computer speakers to be plugged straight in. They are also conveniently small connectors. However they are not especially good quality connectors for analogue audio use, being prone to poor contact affecting the sound, and aren't very robust. Where best quality is more important than convenience, a better connector type is preferable, such as phono sockets.
A series resistor should be fitted between the distribution cable and every socket on the cable (one resistor on each of the L & R lines) for several reasons:
- Resistors protect the audio feed against faults in receiving equipment and leads
- Resistors increase the number of headphones that can be plugged in at once.
- Resistors prevent a local problem from distorting the feed or frying the feed amplifier
- TRS jack connectors often short during insertion, and resistors protects the feed from being shorted
- Protection against TRS shorting is still needed when no TRS connectors are used on the feed line, as TRS connectors on the other end of a lead plugged into the feed will still cause shorting on insertion.
- Resistors protect the feed amp from various scenarios that would otherwise fry it.
When using a low power feeder amplifier (eg 1 watt), low resistance values should be used at the sockets to avoid significant loss of volume with headphone use. There is no specific ideal value, but something in the region of 15-22 ohms would be good for a typical domestic setup.
See #Multiple feeds for information on resistors connected to the feed amp's output.
When exterior wiring is used, the exterior cable run should be fed by a pair of 15 ohm or so resistors. Now if faults occur on the exterior cable it doesn't stop the rest of the system working.
A small amp of 0.1w to 3 watts is best, preferably the lower end of the power range. More power is counterproductive, and only increases hiss levels.
The amp should be run from a wallwart via a suitable fuse on the dc side. A 1A fuse is fine. The fuse is wise because many small amps have no output device protection, there are many chances to short the distribution cable, and the small power dissipation is concentrated in very tiny dice in the amplifier.
Computer speakers with a headphone socket are about as easy a feed amplifier as is available. They're common, all necesary connections are externally pluggable, and they come with their own wallwart supply. No need to even open them, just plug in. Connecting the 3.5mm jack input to a hifi requires either a 2 phono plug to 3.5mm stereo socket adaptor, or replacing the 3.5mm plug with a pair of phonos. The miserable quality speakers built in to them have no effect on the sound in this use.
Uncased low power amplifiers are available from electronic component suppliers for a few pounds. Its usually necessary to solder the wire connections to the PCB and provide a power supply.
Most small amps behave well enough when run with a higher than designed load impedance (as this system does), but there is one notable exception. The LM386 is a minimum cost 8 pin audio amp chip with sound quality suited to answerphones and novelty devices. The LM386 has a fair bit of distortion, and is easily spotted by the mess it makes of treble frequencies.
An unused stereo hifi amp could be used, but will not give good performance if connected directly, with high hiss levels. This problem can be solved with a voltage divider:
amp + -------+ | R1 | +------ distribution cable + | R2 | amp - -------+------ distribution cable ground
For a 30w/channel (rms) amp, R1 = 33 ohm 5 watt, R2 = 15 ohm 2.5 watt (ignore pmpo power ratings).
when one is used, set bass treble and balance to central position.
Most people have a small amplifier somewhere, sitting in an unused appliance. Computer speakers, radios, in fact anything that can drive speakers will contain an audio amp. Appliances that can drive headphones at loud volume, such as a CDROM, also have a usable small amp inside. Identifying the right places to input signals is often required, and often removing the existing signal input.
Computer speakers are easier, as there is nothing but an amplifier on the circuit board, so no need to work out where to solder inputs and power to. A pair with a headphone socket makes life really easy, as everything just plugs in.
Multiple feed points are not usually used, but can be. One minor modification is required; resistors need to be inserted between each feed amplifier output and the feed line. The value of each resisor should be 8 ohms multiplied by the number of feed amplifiers connected to the line, so 16 ohms for 2 feed amps, or 24 ohms with 3 feed amps. These values aren't standard values, and the next value up is used, ie 18 ohms and 27 ohms.
Adding a mic at the feed points enables the system to be used as a PA system. This can sometimes reduce the amount of run around with kids, and can be used as a basic help call system. To do this the feed amp or something feeding the feed amp needs to have a microphone input. A momentary press switch is used to connect the mic, and preferably a relay is added to also silence the local speaker (to prevent feedback).
Audible notifications can be fed to the system from anywhere in the house. Probably the most useful of these is the door bell. If the bell output is fed to the distribution network, the bell should still have its own local sounder as well, as its sensible to turn off the bell's feed when recording. When signalling devices are connected, running them all off a central power feed makes it easy to switch them all off.
Note that audio feed devices such as doorbells may have the wrong voltage, impedance and ground levels for direct connection to the feed line. Resistors and sometimes capacitors are used to correct this.
when feeding such signals to the line, the signal sources are treated like another feed amplifier, requiring suitable resistors to be added to all feed sources. See #Resistors for details.